AI ECONOMIC SALVATION OR SOCIETAL PERIL? Talk_April 2018.pdf · Accenture Research Bath Royal...
Transcript of AI ECONOMIC SALVATION OR SOCIETAL PERIL? Talk_April 2018.pdf · Accenture Research Bath Royal...
AI – ECONOMIC SALVATION OR
SOCIETAL PERIL?Should we welcome or fear the arrival of artificial intelligence?
Mark PurdyManaging Director, Economic Research
Accenture Research
Bath Royal Literary and Scientific Institution
23 April 2018Image: Kai Stachowiak
https://creativecommons.org/publicdomain/zero/1.0/
Copyright © 2018 Accenture. All rights reserved. 2
INTRODUCTION – SPEAKER
Mark Purdy,
Managing Director - Economic Research
Accenture Research
Copyright © 2018 Accenture. All rights reserved. 3
FOR MANY OF US, OUR FIRST INTRODUCTION TO THE IDEA
OF ARTIFICIAL INTELLIGENCE CAME VIA THE MOVIES
AI in service to the society or AI posing threats to humanity?
TARS, the space robot from ‘Interstellar’ (2014)
Robotics fighting back from ‘I, Robot’ (2002)
Crime forecasting from ‘Minority Report’ (2002)
Baymax, the healthcare robot from ‘Big Hero 6’ (2014)
A supercomputer deeming mankind a threat, Singularity (2017)
Machines taking control of Earth, ‘the Matrix’ (1999)
v.s.
Copyright © 2018 Accenture. All rights reserved. 4
HUMAN-LEVEL AI IS STILL A LONG WAY OFF
ARTIFICIAL GENERAL INTELLIGENCEARTIFICIAL NARROW INTELLIGENCE
Understand abstract concepts
Explain why
Be creative like children
Tell right from wrong
Have emotions
Beat Go world champions
Read facial expressions
Write music
Diagnose mental disorders
Comfort earthquake survivors
MORAVEC’S PARADOX
What AI can and can’t do
Copyright © 2018 Accenture. All rights reserved. 5
WHAT IS ARTIFICIAL INTELLIGENCE
Act
Sense
Comprehend
Computer Vision
Audio Processing
Natural Language Processing
Knowledge Representation
Machine Learning
Expert Systems
AI TECHNOLOGIES ILLUSTRATIVE SOLUTIONS
Virtual Agents
Identity Analytics
Cognitive Robotics
Speech Analytics
Recommendation
Systems
Data Visualization
Copyright © 2018 Accenture. All rights reserved. 6
AI HAS A LONG HISTORY OF SUMMERS AND WINTERS
19901950 1960 1970 1980 2000 2010
The 1st boom
• Reasoning as research
• Natural language
• WABOT-1-world’s first full-scale
intelligent humanoid robot
The birth of AI
• Cybernetics - thinking machines
• The Turning test
• Dartmouth conference
• Symbolic reasoning
The 1st AI winter
• Limited computer power
• Limited database capacity
• Limited networking capabilities
• Real-world problems are complicated
The 2nd boom
• The rise of expert systems
• The knowledge revolution
• Neural networks make a
comeback
The 2nd AI winter
• A sudden collapse of specialized AI
hardware vendors
• Nouvelle AI and embodied reason
2000-present
• Deep learning, big data and
artificial general intelligence
Copyright © 2018 Accenture. All rights reserved. 7
RECENTLY IT HAS BEEN UNDERGOING ANOTHER BOOM
AI Startup investment growth is outpacing that of
the overall startup market (2011-2016)
Global AI vs overall startup investment markets ($)
Overall market startup
investmentAI startup investment
54% 16%vs.
AI investments ($)
9X 6X
AI deals (#)
By 2016, 11% of all global startup deals were AI-oriented, up from 2% in 2010
0
2,000
4,000
6,000
8,000
10,000
Priority Year
2007
Num
be
r o
f P
rio
rity
Pa
ten
ts
2006 201020092008 2011 2012 2013 2014 2015c*
+24%
Filed Priority Patents – Artificial Intelligence
named filings and related technology filings
AI keywords (atleast one AI technology directly named in patent filing)
AI smartsearch (AI related tech filings)
*Note: For year 2015, only 6 months data is published and full year data is assumed at a
factor of two
Both AI startup investment and patent publication on AI have accelerated since 2010
Copyright © 2018 Accenture. All rights reserved. 8
BECAUSE OF THREE DRIVING FORCES
Rapid increases in
computational powerExplosion in data that is
digitized
Huge fall in cost of storing data –
near zero marginal cost
One dollar’s worth of computer power, 1980-2010
Billions of computations per second
(log scale)
Source: The Brookings Institution
Hard drive cost per gigabyte, 1980-2015
Global installed base of IoT units (number of units, billions)
0
5
10
15
20
25
2015
2016
2017f
2018f
2019f
2020f
Business Consumer
Source: Gartner, Forecast: Internet of Things - Endpoints and Associated Services, Worldwide, 2016
Source: mkomo.com
1,000,000
0.01
0.10
1,000
100,000
10,000
100
10
1
USD (log scale)
1980 1985 1990 1995 2000 2005 2010 2015
Copyright © 2018 Accenture. All rights reserved. 9
How people feel about AI?
• Public discussion has exploded since 2009
• In general, AI has had consistently more optimistic than
pessimistic coverage over time
THE RAPID ADVANCE OF AI HAS LED TO HEATED DEBATES
AROUND ITS POTENTIAL THREATS AND OPPORTUNITIES
Optimistic
articles
Pessimistic
articles
Articles on AI
Source: Long-term trends in the public perception of AI, Fast & Horvitz, 2016, Association for
the Advancement of Artificial Intelligence
AI in the news: The potential economic benefits and risks
10
COULD AI BE AN ECONOMIC
SALVATION?
Copyright © 2018 Accenture. All rights reserved. 11
GLOBAL ECONOMIC GROWTH IS WEAKENING
-6
-4
-2
0
2
4
6
8
10
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
Developed economies Emerging economies World
GDP growth (real, % y/y)
Source: Oxford Economics
Copyright © 2018 Accenture. All rights reserved. 1212
LOOKING BACK, GDP GROWTH HAS STEADILY SLOWED IN
MANY LARGE ECONOMIES SINCE THE 1980S
1980s 1990s 2000s 2010s
Japan United States United Kingdom France Italy Germany
3.0
2.1
1.1 1.1
NB: Data points across the dashed lines indicate the average for the six countries.
Source: Oxford Economics
Real GDP growth (%, annual average over the period)
Copyright © 2018 Accenture. All rights reserved. 13
THIS IS BECAUSE THE TRADITIONAL FACTORS OF
PRODUCTION ARE UNDER PRESSURE
Capital efficiency: The marginal capital efficiency rate has steadily dropped over a 50-year period.
Marginal capital efficiency (%, 6-year moving average)
0
5
10
15
20
25
30
35
40
1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 2011 2016
Labour:As populations age and birth rates slow, fewer people are available to pick up the slack in the workforce.
Working age population
(%, annual average growth over the period)
0.9
0.6
0.4
0.1
-0.1
-0.69
0.30.2
-0.05
-0.3
-0.7-0.73
UnitedStates
UnitedKingdom
France Italy Germany Japan
2000-2015 2016-2030
Japan United States Germany United Kingdom
Copyright © 2018 Accenture. All rights reserved. 1414
…AND PRODUCTIVITY GROWTH IS TRENDING DOWNWARD
Total Factor Productivity (%, annual average over the period)
2.1
1.00.6
-0.1
0.7
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
1990-19941995-19992000-20042005-20092010-2014
0.4
-0.2
0.7
-0.9
-0.2
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
1990-19941995-19992000-20042005-20092010-2014
00.5 0.5
-0.7
0.1
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
1990-19941995-19992000-20042005-20092010-2014
Source: The Conference Board, Total Economic Database
0.5 0.71.1
0.040.5
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
1990-19941995-19992000-20042005-20092010-2014
0.3 0.1
0.9
-0.4
0.8
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
1990-19941995-19992000-20042005-20092010-2014
0.50.3
-0.5-1.3
-0.05
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
1990-19941995-19992000-20042005-20092010-2014
United States Japan Germany
France United
Kingdom
Italy
A key measure of how well an economy uses its existing capital and people is “total factor productivity” (TFP).
Data show a weakening of TFP, especially in the past 10 years.
Copyright © 2018 Accenture. All rights reserved. 15
SO ARE WE EXPERIENCING THE END OF GROWTH AND
PROSPERITY?
Copyright © 2018 Accenture. All rights reserved. 1616
AI CAN BE A POWERFUL REMEDY FOR SLOWING GROWTH
Source: Accenture analysisindicates the change in that factor.NB:Δ
AI has the potential to overcome the physical limitations of capital and labour and open up new sources of value
and growth
Capital Labour TFP
Growth
Capital Labour TFP AI
Traditional Growth
Model
Adapted Growth Model
Δ Δ
ΔΔ Δ
Copyright © 2018 Accenture. All rights reserved. 17
AI CAN BOOST NATIONAL ECONOMIC GROWTH VIA THREE
CHANNELS
AugmentationIntelligent
Automation
Innovation
Diffusion
1 2 3
Copyright © 2018 Accenture. All rights reserved. 1818
AI PROMISES TO BOOST NATIONAL ECONOMIC GROWTH
2.6
2.12.5
1.7 1.61.4 1.4
1.7
0.8
1.6 1.7
1.0
4.6
4.13.9
3.63.2 3.0 3.0 2.9
2.7 2.7 2.5
1.8
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Baseline AI steady state
Real gross value added (GVA)
(%, growth)
AI has the potential to double annual growth rates in the countries we analyzed in terms of gross value added
(a close approximation of GDP).
Source: Accenture and Frontier Economics
19
CONCERNS AND FEARS AROUND AI
ARE ON THE RISE
Copyright © 2018 Accenture. All rights reserved. 20
AI IS ON THE CUSP OF TRANSFORMING EVERY ASPECT OF
HOW WE DO OUR JOBS AND LIVE OUR LIVES
HOW WE LIVEHOW WE WORK
• Labour market
disruption
• Future of jobs
• Erosion of individual
identity
• …
• Data privacy
• Bias and fairness
• Growing inequality
• Post-work society
• …
Copyright © 2018 Accenture. All rights reserved. 21
CONCERNS THAT TECHNOLOGY WILL DESTROY OUR JOBS
ARE NOT NEW
2011
David Ricardo,
The Principles of
Political Economy
and Taxation“…the substitution of machinery
for human labour, is often very
injurious to the interests of the
class of labourers.”
Karl Marx, Capital
“…this effect of machinery…
represented to be a
compensation to the working
class, is on the contrary a most
frightful scourge.”
John M. Keynes,
Economic Possibilities
for our Grandchildren“We are being afflicted with a new disease...
namely, technological unemployment.
This means unemployment due to our
discovery of means of economising the use
of labour outrunning the pace at which we
can find new uses for labour.”
1821 1887 1930 1952 20131989
Kurt Vonnegut,
Player PianoA novel that depicts a
world dominated by a
supercomputer and run
completely by machines
Paul Samuelson,
Ricardo was Right!Frey & Osborne,
The Future of
Employment:
How Susceptible
are Jobs to
Computerisation?
McAfee &
Brynjolfsson,
Race against
the Machine
Copyright © 2018 Accenture. All rights reserved. 22
HISTORICALLY, TECHNOLOGY HAS AFFECTED LABOUR
MARKET TRANSITIONS IN FOUR DIFFERENT WAYS
Substitution
EffectComplementary
Effect
Income
Effect
Price
Effect
Industrial robots took over
manual tasks such as welding,
painting and assembling
Computers improved efficiency
and created new jobs such as
computer-aided designers
Full-time earnings for British
workers more than doubled in
the seventy years after 1780
Transportation innovations and
development sharply pushed
down the cost of travelling
Copyright © 2018 Accenture. All rights reserved. 23
MH
NEW EMPLOYEE ROLES AND NOVEL TYPES OF WORKING
RELATIONSHIPS ARE REQUIRED
Human-only
activity
Machine-only activity
Humans complement
machinesAI gives
humans
superpower
Human and machine
hybrid activities
Le
ad
Em
pa
thiz
e
Cre
ate
Ju
dg
e
Tra
in
Su
sta
in
Exp
lain
Am
plif
y
Inte
ract
Em
bo
dy
Tra
nsa
ct
Ite
rate
Pre
dic
t
Ad
ap
t
The missing middle: new jobs, roles, and skills will be required as organizations use AI to augment and
transform the way they work
Source: Human + Machine: Reimagining Work in the Age of AI, Daugherty & Wilson, Harvard Business Review Press, 2018
Copyright © 2018 Accenture. All rights reserved. 24
AI BECOMES DEEPLY INGRAINED IN OUR EVERYDAY LIVES,
YET REAL CHALLENGES REMAIN REGARDING PRIVACY,
TRUST AND EXISTING BIASES
Source: Accenture Dynamic Consumers, 2017
Consumers want to be more engaged in managing
their data
Slaves to the algorithms?
of consumers believe it is important
to be able to review and control their
personal data online. 87%
3/4however, do not find it easy to
manage that data.
• Reinforcing existing stereotypes, social and
cultural segregation and exclusion
• Personalized recommendation leading to a
‘filter bubble”?
Copyright © 2018 Accenture. All rights reserved. 25
BUT AI COULD BE THE PIVOTAL TECHNOLOGY IN
ADDRESSING THE GRAND CHALLENGES OF HUMANITY
Inclusive education and personalized
learning
Empowering the disabled
Health management and disease
diagnosis
Fight against world hunger
Fight against crime
Urban traffic managementThe AI traffic system in Pittsburgh- Surtrac which can
detect traffic and allow traffic lights to adapt to real-
time conditions has helped cut travel time by 25%
Knewton, an AI-enabled adaptive learning platform leverages machine learning to personalize content to each user to improve academic outcomes.
Seeing AI, a free mobile application designed to support people with visual impairments by narrating the world around them.
DreaMed Diabetes, an Israel-based start-up, developed an algorithm which continuously monitors patients’ glucose levels and defines precisely when and how to adjust insulin levels.
Stanford University has instituted an AI program that uses machine learning to predict crop yield. The goal is to provide planters with information allowing them to make more thorough planting and food reservation decisions.
The Murder Accountability Project (MAP) has developed an online serial-detection algorithm. Established in 2015, this serial-detection system is designed to identify areas with significantly low clearance rates through the recognition of the homicidal patterns of individual serial killers.
26
How to realize AI’s promise while
controlling the peril
It all depends on how we manage
the transition to an era of AI
Copyright © 2018 Accenture. All rights reserved. 27
FORGING A DIGITAL NEW DEAL THAT ALLIES ECONOMIC
PROSPERITY WITH SOCIAL PROGRESS
Copyright © 2018 Accenture. All rights reserved.
• Imagine the future
• Address grand challenges
• Catalytic investment
• Regulatory innovation
• Reskilling for Jobs of the
Future
• Social and Technical skills
• Transition paths for affected
workers
• Life-long education system
• Responsible AI
• Missing Middle skills
• Regional & sectoral
adjustment
• Grassroots support
A digital 'New Deal’ - Three priorities
Big
Ambition
Big
Inclusion
Big
Skills Agenda
28
ANNEX
Copyright © 2018 Accenture. All rights reserved. 29
READY FOR HUMAN-MACHINE COLLABORATION ?
It will be important/very important to learn new
skills to work with intelligent technologies in the
next 3 to 5 years.
Copyright © 2018 Accenture. All rights reserved.
Employers underestimate the willingness of employees to acquire the relevant skills
62%
I believe intelligent technologies will create
opportunities for my work.
67%
ONLY 3%of executives say they intend to significantly increase
investment in training and reskilling programs in the
next three years.
Sample: More than 1,200 CXOs, more than 14,000 workers, 12 industriesSource: Accenture, Reworking the revolution, 2017
Copyright © 2018 Accenture. All rights reserved. 30
AI WILL BE BOTH A DISRUPTING AND ENABLING FORCE IN
THE WORKPLACE
Over the years, technology has made some jobs redundant,
but also led to the creation of new jobs and occupations:
Source: OECD, OECD Employment Outlook 2017
High-skill
occupations
Middle-skill
occupations
Low-skill
occupations
7.6
-9.5
1.9
Copyright © 2018 Accenture. All rights reserved.
Change of share of employment in the OECD area, 1995-
2015 (percentage points)• Free us up from 3 “D’s” – Dull, Dirty, Dangerous
But the impact of AI is more than automation; It’s
more about workplace augmentation
Legal analysis
AI surgical robotsVirtual assistants
• Make us more productive and innovative
Copyright © 2018 Accenture. All rights reserved. 31
AI RISKS EXACERBATING INCOME INEQUALITY AND
POLARIZATION OF SOCIETIES ACROSS THE WORLD
Technological progress has contributed the most to
widening income inequality across the globe from mid-
1980s to mid 2000s
(average annual percent change)
Source: IMF
-0.25 0 0.25 0.5
Change in inequality
Contribution of trade andfinancial globalization
Contribution of technology
Contribution of other
…And a new digital divide is looming on the
horizon
0
20
40
60
80
100
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
High Income Countries Middle Income Countries
Low Income Countries World
Internet penetration (%), 2000-2015
Source: OECD
Copyright © 2018 Accenture. All rights reserved. 32
LOOKING AHEAD, AI WILL POSE NEW CHALLENGES TO BOTH
REGULATORS AND BUSINESSES
Limits of Taylorism
Universal basic incomes
Robot taxes
Digital inclusivenessLearn with the machine
Transition paths to
new jobs